Automated deep learning may provide an opportunity to greatly extend the use of artificial intelligence (AI) in healthcare, according to an article published in Lancet Digital Health. This would allow medical professionals that aren’t experts in AI to be able to build AI algorithms, allowing a much wider range of applications to be explored. In the future, these approaches could offer huge potential benefits for patients by allowing earlier detection and treatment of disease.
Researchers at a number of institutions, led by Moorfields Eye Hospital NHS Foundation Trust and UCL Institute of Ophthalmology, used Google AutoML, a machine-learning algorithm that learns to build other machine-learning systems (‘AI that can build AI’), to build an algorithm to analyse medical images and provide diagnoses. For simple classification tasks, they were able to demonstrate a diagnostic performance similar to that of many state-of-the-art AI systems.
However, the authors caution that, although this technology provides possibilities for clinicians who aren’t experts in AI, further work will be required before these approaches can be applied in real world clinical practice. The researchers expressed the need for deep learning experts and clinicians to collaborate to ensure AI is used correctly in clinical practice and highlighted the need for regulatory guidelines in this area.
The researchers used five publicly available medical image datasets, including chest x-rays, skin lesions, and two types of eye scan. The images were labelled and entered into the AutoML programme, which created an automated algorithm that was used to classify the diseases in the images using AI. The authors evaluated how accurately the model predicted the eye conditions on the medical images.
Pearse Keane, consultant ophthalmologist at Moorfields Eye Hospital and NIHR clinician scientist said:
“At present, the development of AI systems requires highly specialised technical expertise. If this technology can be used more widely – in particular by healthcare professionals without computer programming experience – it will really speed up the development of these systems with the potential for significant patient benefits.
“While the derivation of classification models without requiring a deep understanding of the mathematical, statistical and programming principles is attractive, comparable performance to expertly-designed models is currently limited to more simple classification tasks. The process needs refining and regulation, but our results show promise for the future expansion of AI in medical diagnosis.”
Ailish Murray, director of grants and research a Moorfields Eye Charity, said:
“It’s still early days but these results are really encouraging when we consider their potential to bring clinical benefits to a wider range of patients. We’re excited to see what our continued funding of ground-breaking AI research at Moorfields will lead to next.”
This research was funded by Moorfields Eye Charity and the National Institute for Health Research.
Was this information useful? Please rate the page.